INTERNET INTRUSION DETECTING ALGORITHM BASED ON THE CULTURE ALGORITHM AND K-MEAN CLUSTERING

In order to overcome the shortcoming of the traditional K-mean clustering algorithm which greatly depends on the choice of the initial value and thus is easy to stuck in the local extremum, we have proposed an intrusion detecting algorithm that combines the culture algorithm (CA) with the K-mean clustering algorithm. The two-layer structure characteristic of the culture algorithm enables it to show a good global optimizing performance in the search process, which can overcome effectively the two major shortcomings of the traditional K-mean algorithm. In KDD Cup1999 standard intrusion detecting data set, we have done a simulation test, and the test result has shown that CAKM algorithm is better than the intrusion detecting algorithm based on K-mean clustering algorithm and based on genetic algorithm, and can enhance the detection rate of the intrusion detecting system and lower its misreport rate to a certain extent.